BAIWL: Blacklisting Approach to Improve Wireless Sensor Network Lifetime

Smart anomaly detection in sensor systems: A multi-perspective review

“…Moreover, authors in [111] proposed a system design for federated learning, which is a kind of distributed machine learning approach, to enable model training using decentralized data from devices such as mobile phones. ⋅ Distributed ML techniques [109], [110] ⋅ Efficient fusion strategies [111] Acceleration Circuitry Limits ⋅ FPGA-based optimization [112], [113] ⋅ Algorithm complexity reduction [113], [114], [115] Energy Efficiency Power Limits ⋅ Efficient routing protocols [116], [117], [118], [119] ⋅ Smart topologies [117], [118] ⋅ Lightweight operating systems [120] Security Computational Limits ⋅ Distributed security mechanisms [121], [122] ⋅ Novel keys distribution schemes [123] ⋅ False data injection [124], [125] Sensors Softwarization Special-purpose architectures ⋅ NFV/SDN for sensors [126], [127], [128], [129] ⋅ Integration with the 5G paradigm [130], [131] Architectural Models Intelligence Distribution ⋅ Hierarchical Learning [13] ⋅ Hybrid models [132], [133] Data Heterogeneity Aggregation ⋅ Ad-hoc middle-wares to homogenize data [134], [135] ⋅ New fusion strategies accounting for data variability [134], [136] Both the distributed and the decentralized ML approaches present potential alternatives to centralized ML, particularly in the context of hybrid computing models. At the same time, new data-fusion strategies are required to consider the heterogeneity of data sources and handle the "missing data" problem [138], which is particul...…”

“…Remarkably, the authors in [117] propose an approach aimed at forbidding low energy nodes to be forwarder nodes when the residual energy is lower than a specific threshold. Again, an interesting extension of RPL to mobility nodes (useful, for instance, within the automotive environment) is advanced in [118] where, by exploiting additional fields of RPL control packets, it is possible to advice a mobile sensor node for maintaining the energy by reducing network overheads.…”